PipeWeaver: Addressing Data Dynamicity in Large Multimodal Model Training with Dynamic Interleaved Pipeline

• URL: http://arxiv.org/abs/2504.14145v1
• Date: Sat, 19 Apr 2025 02:30:11 GMT
• Title: PipeWeaver: Addressing Data Dynamicity in Large Multimodal Model Training with Dynamic Interleaved Pipeline
• Authors: Zhenliang Xue, Hanpeng Hu, Xing Chen, Yimin Jiang, Yixin Song, Zeyu Mi, Yibo Zhu, Daxin Jiang, Yubin Xia, Haibo Chen,
• Abstract summary: PipeWeaver is a dynamic pipeline scheduling framework for large multimodal models (LMMs)<n>It addresses issues of LMM training with two techniques: adaptive modality-aware partitioning and efficient pipeline schedule search.<n>Experiments show that PipeWeaver can enhance LMM training efficiency by up to 97.3% compared to state-of-the-art systems.
• Score: 39.56060323129395
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: Large multimodal models (LMMs) have demonstrated excellent capabilities in both understanding and generation tasks with various modalities. While these models can accept flexible combinations of input data, their training efficiency suffers from two major issues: pipeline stage imbalance caused by heterogeneous model architectures, and training data dynamicity stemming from the diversity of multimodal data. In this paper, we present PipeWeaver, a dynamic pipeline scheduling framework designed for LMM training. The core of PipeWeaver is dynamic interleaved pipeline, which searches for pipeline schedules dynamically tailored to current training batches. PipeWeaver addresses issues of LMM training with two techniques: adaptive modality-aware partitioning and efficient pipeline schedule search within a hierarchical schedule space. Meanwhile, PipeWeaver utilizes SEMU (Step Emulator), a training simulator for multimodal models, for accurate performance estimations, accelerated by spatial-temporal subgraph reuse to improve search efficiency. Experiments show that PipeWeaver can enhance LMM training efficiency by up to 97.3% compared to state-of-the-art systems, and demonstrate excellent adaptivity to LMM training's data dynamicity.

Related papers

• Demystifying Workload Imbalances in Large Transformer Model Training over Variable-length Sequences [31.232756326457277]
  We develop Hydraulis, which jointly optimize the parallel strategies and data assignment. Empirical results demonstrate that Hydraulis outperforms existing systems by 1.32-2.66 times.
  arXiv  Detail & Related papers  (2024-12-10T20:01:53Z)
• EPS-MoE: Expert Pipeline Scheduler for Cost-Efficient MoE Inference [49.94169109038806]
  This paper introduces EPS-MoE, a novel expert pipeline scheduler for MoE that surpasses the existing parallelism schemes.<n>Our results demonstrate at most 52.4% improvement in prefill throughput compared to existing parallel inference methods.
  arXiv  Detail & Related papers  (2024-10-16T05:17:49Z)
• Spindle: Efficient Distributed Training of Multi-Task Large Models via Wavefront Scheduling [35.06717005729781]
  Spindle is a new training system tailored for resource-efficient training of multi-task (MT) multi-modal (MM) models via wavefront scheduling.<n>Experiments demonstrate the superior performance and efficiency of Spindle, with speedup ratio up to 71% compared to state-of-the-art training systems.
  arXiv  Detail & Related papers  (2024-09-05T09:10:40Z)
• Dynamic Data Mixing Maximizes Instruction Tuning for Mixture-of-Experts [20.202031878825153]
  We propose a novel dynamic data mixture for MoE instruction tuning. Inspired by MoE's token routing preference, we build dataset-level representations and then capture the subtle differences among datasets. Results on two MoE models demonstrate the effectiveness of our approach on both downstream knowledge & reasoning tasks and open-ended queries.
  arXiv  Detail & Related papers  (2024-06-17T06:47:03Z)
• DynaPipe: Optimizing Multi-task Training through Dynamic Pipelines [15.332562681746081]
  This paper proposes a dynamic micro-batching approach to tackle sequence length variation and enable efficient multi-task model training. We optimize micro-batch construction using a dynamic programming-based approach, and handle micro-batch execution time variation through dynamic pipeline and communication scheduling.
  arXiv  Detail & Related papers  (2023-11-17T09:48:45Z)
• Efficient Multimodal Fusion via Interactive Prompting [62.08292938484994]
  Large-scale pre-training has brought unimodal fields such as computer vision and natural language processing to a new era. We propose an efficient and flexible multimodal fusion method, namely PMF, tailored for fusing unimodally pre-trained transformers.
  arXiv  Detail & Related papers  (2023-04-13T07:31:51Z)
• FlexMoE: Scaling Large-scale Sparse Pre-trained Model Training via Dynamic Device Placement [19.639936387834677]
  Mixture-of-Experts (MoEs) are becoming more popular and have demonstrated impressive pretraining scalability in various downstream tasks. MoEs are becoming a new data analytics paradigm in the data life cycle and suffering from unique challenges at scales, complexities, and granularities never before possible. In this paper, we propose a novel DNN training framework, FlexMoE, which systematically and transparently address the inefficiency caused by dynamic dataflow.
  arXiv  Detail & Related papers  (2023-04-08T07:34:26Z)
• Adapted Multimodal BERT with Layer-wise Fusion for Sentiment Analysis [84.12658971655253]
  We propose Adapted Multimodal BERT, a BERT-based architecture for multimodal tasks. adapter adjusts the pretrained language model for the task at hand, while the fusion layers perform task-specific, layer-wise fusion of audio-visual information with textual BERT representations. In our ablations we see that this approach leads to efficient models, that can outperform their fine-tuned counterparts and are robust to input noise.
  arXiv  Detail & Related papers  (2022-12-01T17:31:42Z)
• Where Is My Training Bottleneck? Hidden Trade-Offs in Deep Learning Preprocessing Pipelines [77.45213180689952]
  Preprocessing pipelines in deep learning aim to provide sufficient data throughput to keep the training processes busy. We introduce a new perspective on efficiently preparing datasets for end-to-end deep learning pipelines. We obtain an increased throughput of 3x to 13x compared to an untuned system.
  arXiv  Detail & Related papers  (2022-02-17T14:31:58Z)
• TeraPipe: Token-Level Pipeline Parallelism for Training Large-Scale Language Models [60.23234205219347]
  TeraPipe is a high-performance token-level pipeline parallel algorithm for synchronous model-parallel training of Transformer-based language models. We show that TeraPipe can speed up the training by 5.0x for the largest GPT-3 model with 175 billion parameters on an AWS cluster.
  arXiv  Detail & Related papers  (2021-02-16T07:34:32Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.